Rackscale applications like Hadoop, Spark, Cassandra and others count on using commodity storage that is typically internally available to the node processing the data. The idea is to reduce storage costs and network complexity. The problem is these designs create…
